Fluid overload, de-resuscitation, and outcomes in critically ill or injured patients: a

systematic review with suggestions for clinical practice

Abstract
Background: Sepsis is associated with generalised endothelial injury and capillary leak and has
traditionally been treated with large volume fluid resuscitation. Some patients with sepsis
will accumulate bodily fluids. The aim of this study was to systematically review the association
between a positive fluid balance/fluid overload and outcomes in critically ill adults, and to
determine whether interventions aimed at reducing fluid balance may be linked with improved
outcomes.
Methods: We searched MEDLINE, PubMed, EMBASE, Web of Science, The Cochrane Database,
clinical trials registries, and bibliographies of included articles. Two authors independently
reviewed citations and selected studies examining the association between fluid balance and
outcomes or where the intervention was any strategy or protocol that attempted to obtain a negative
or neutral cumulative fluid balance after the third day of intensive care compared to usual care.
The primary outcomes of interest were the incidence of IAH and mortality.
Results: Among all identified citations, one individual patient meta-analysis, 11 randomised
controlled clinical trials, seven interventional studies, 24 observational studies, and four case series
met the inclusion criteria. Altogether, 19,902 critically ill patients were studied. The cumulative
fluid balance after one week of ICU stay was 4.4 L more positive in non-survivors compared to
survivors. A restrictive fluid management strategy resulted in a less positive cumulative fluid
balance of 5.6 L compared to controls after one week of ICU stay. A restrictive fluid management
was associated with a lower mortality compared to patients treated with a more liberal fluid
management strategy (24.7% vs 33.2%; OR, 0.42; 95% CI 0.32−0.55; P < 0.0001). Patients with
intra-abdominal hypertension (IAH) had a more positive cumulative fluid balance of 3.4 L after
one week of ICU stay. Interventions to decrease fluid balance resulted in a decrease in intra-
abdominal pressure (IAP): an average total body fluid removal of 4.9 L resulted in a drop in IAP
from 19.3 ± 9.1 mm Hg to 11.5 ± 3.9 mm Hg.
Conclusions: A positive cumulative fluid balance is associated with IAH and worse outcomes.
Interventions to limit the development of a positive cumulative fluid balance are associated with
improved outcomes. In patients not transgressing spontaneously from the Ebb to Flow phases of
shock, late conservative fluid management and late goal directed fluid removal (de-resuscitation)
should be considered.
Key words: adults, critical care, fluid therapy, sepsis, capillary leak, fluid overload, goal directed,
resuscitation, conservative fluid management, deresuscitation, ROSE conceptual model,
monitoring.

The administration of intravenous fluids is widely regarded as the first step in the
resuscitation of critically ill and injured patients who have evidence of impaired organ
perfusion [1−3]. The Surviving Sepsis Campaign recommends “aggressive fluid resuscitation
during the first 24 hours of management” [4]. The purpose of fluid resuscitation is to increase
venous return and stroke volume [5]. Fluid administration increases the stressed blood volume,
increasing the gradient between the mean systemic filling pressure (MSFP) and right atrial pressure
(CVP), thereby increasing venous return [6-8]. In patients who are on the ascending limb of
the Frank-Starling limb, the increased venous return results in an increase in stroke volume and
cardiac index [5].
Despite the above, clinical studies have consistently demonstrated that less than 50% of
haemodynamically unstable patients are fluid responders, as defined by an increase in stroke
volume of 10−15% following a fluid challenge [5]. Fluid administration serves no useful purpose
in those patients whose stroke volume fails to increase following a fluid challenge (non-
responders). In these patients, f luid administration may even be harmful. Furthermore, due to the
redistribution of fluid, the haemodynamic response in fluid responders is short lived with the stroke
volume returning to baseline 30 to 60 minutes following the initial fluid challenge [9]. In
healthy individuals, 85% of an infused bolus of crystalloid has been reported to redistribute into
the interstital space after four hours [10]. In critically ill patients with endothelial injury and leaky
capillaries, less than 5% of a fluid bolus remains intravascular after 90 minutes [11]. In the Rivers’
Early Goal Directed Therapy (EGDT) study, 4.9 L of crystalloid were given in the first six hours
and 13.4 L in the first 72 hours [12]. The Surviving Sepsis Campaign recommends “aggressive
fluid resuscitation during the first 24 hours of management” [4].
Large volume fluid resuscitation results in severe tissue oedema and clinical signs of
volume overload [13]. Tissue oedema impairs oxygen and metabolite diffusion, distorts tissue
architecture, impedes capillary blood flow and lymphatic drainage, and disturbs cell-cell
interactions [13]. These effects are pronounced in encapsulated organs, such as the liver and
kidneys, which lack the capacity to accommodate additional volume without an increase in
interstitial pressure, resulting in compromised organ blood f low [14]. Furthermore, large
volume resuscitation increases intra-abdominal pressure (IAP), which further compromises renal
and hepatic perfusion. As such, capillary leak significantly contributes to the genesis of intra-
abdominal hypertension (IAH) and the abdominal compartment syndrome (ACS) [15−19). Kelm
et al. [20] demonstrated that 67% of patients resuscitated by means of the EGDT protocol had
clinical evidence of fluid overload after 24 hours, with 48% of patients having persistent features
of fluid overload by the third hospital day.
Multiple studies have demonstrated that a positive f luid balance is independently
associated with impaired organ function and an increased risk of death [14, 15, 21−29]. This was
recently demonstrated in an elegant study by Murphy et al. [21]. Conversely, achievement of a
negative fluid balance is associated with improved organ function and survival [30, 31]. This has
been referred to as the Ebb and Flow phases of shock. The Ebb phase was characterised by
Cuthbertson in 1932 as: “Ashen faces, a thready pulse and cold clammy extremities…”, while
during the Flow phase “the patient warms up, cardiac output increases and the surgical team
relaxes…” [25]. Recent data suggests that many patients do not enter the flow phase spontaneously
and in order to avoid a positive cumulative fluid balance with the associated organ oedema and
organ dysfunction, they may require therapeutic interventions [32]. However, it remains largely
unknown whether strategies that target a neutral or even negative fluid balance after the initial
resuscitative phase are associated with improved clinical outcomes in humans.
Goal-directed therapy has become ubiquitous, where the goal of resuscitation is the rapid
reversal of shock and hypoperfusion within a few hours. The Surviving Sepsis Campaign
Guidelines focus on the initial resuscitation but fail to provide information on the assessment of
volume overload or when and how to perform de-resuscitation [4]. Furthermore, the central venous
pressure (CVP) provides little useful data as to the patient’s overall volume status and the need for
de-resuscitation. The EV1000/VolumeView (Edwards Lifesciences, Irvine, CA, USA) and PiCCO
(Pulsion Medical Systems, Munich, Germany) devices allow, besides measurement of cardiac
output as well as other parameters such as the global end-diastolic volume index (GEDVI) and
extravascular lung water index (EVLWI) which provide useful information on volume status and
tissue oedema [33–35]. These devices are helpful when faced with a therapeutic conflict, a
situation where each of the possible therapeutic decisions carries some potential harm, with
the clinician supporting the organ that carries the highest danger of harming the patient [36, 37].
In high-risk patients, decisions regarding f luid administration should therefore be done within the
context of a therapeutic conflict.
The aim of this study was to systematically review the association between a positive fluid
balance/fluid overload and outcomes in critically ill adults and to determine whether interventions
aimed at reducing fluid balance may be linked with improved outcomes.
In the discussion we will focus on the available literature with regards to fluid overload
and a positive cumulative f luid balance in relation to morbidity (e.g. IAH) and mortality and
how to deal with it at the bedside. We review de-resuscitation: what, why, when and how?
DEFINITIONS
In this section we will define ‘de-resuscitation’ and suggest some definitions with regard
to fluid management and f luid balance partially based on a conceptual model [38-40].
Resuscitation fluids.
Resuscitation fluids are used to correct an intravascular volume deficit or acute
hypovolemia. Over the last three decades, there has been much debate over the use of colloids vs
crystalloids [41]. However, recent clinical trials suggest that colloids have a limited role in fluid
resuscitation [42−44]. More recently the issue has involved the use of normal saline vs balanced
salt solutions, with data suggesting improved outcomes with balanced salt solutions [45, 46].
Maintenance fluids. Maintenance solutions are specifically given to provide the patient’s
daily basal requirements of water and electrolytes.
Replacement fluids. Replacement solutions are prescribed to correct existing or
developing deficits that cannot be compensated by oral intake, as seen in situations in which f luids
are lost via drains or stomata, fistulas, fever, polyuria and open wounds (including evaporation
during surgery or burns) among others.
Global Increased Permeability Syndrome (GIPS). GIPS is characterised by high capillary
leak index (CLI, expressed as the ratio of CRP over albumin), excess interstitial fluid and persistent
high extravascular lung water index (EVLWI), no late conservative fluid management (LCFM)
achievement, and progression to organ failure [32]. GIPS represents a ‘third hit’ following the
acute injury (first hit) with progression to multi-organ dysfunction syndrome-MODS (second hit)
[47]. The third hit may develop in patients who do not enter the Flow phase spontaneously.
Successful response to acute inflammatory insult tends to be characterised by a crucial turning
point on day three. The evolution of cytokines and other pro-inflammatory mediators on the third
day after shock initiation allows healing of the microcirculatory disruptions and ‘closure’ of
capillary leakage. This interpretation is supported by observations demonstrating normalisation of
microcirculatory blood flow on day three in patients with abdominal sepsis [48]. Further, lower
EVLWI and pulmonary vascular permeability indices (PVPI) at day three of shock have been
shown to correlate with better survival [49]. In these patients, excess fluid administration results
in oedema formation, progression of organ failure and worse outcome. Therefore, as soon as
haemodynamics allow, early transition to conservative fluid management and even fluid removal
on the basis of an EVLWI-guided protocol is mandated (late goal directed fluid removal) [21, 32,
47].
Polycompartment syndrome. The recent consensus definitions of the World Society on the
Abdominal Compartment Syndrome (WSACS, www.wsacs.org) defined polycompartment as a
condition where two or more anatomical compartments have elevated compartmental pressures
[50]. As a result of capillary leak and impaired flow phase, overzealous administration of
unnecessary fluids in the GIPS phase will lead to gross f luid overload and tissue oedema.
Interstitial oedema increases the pressure in all four interconnected major body compartments:
head, chest, abdomen, and extremities. As a result, the venous resistance of organs within
compartments increases and perfusion pressure decreases contributing to progression of organ
failure. As different compartments interact and reciprocally transmit compartment pressures, the
concept of polycompartment syndrome was suggested [51-53]. The abdomen plays a central role
in GIPS and the polycompartment syndrome, as positive fluid balances are a known risk factor for
secondary IAH which in turn is associated with deleterious effects on other compartments and
organ functions and may eventually lead to abdominal compartment syndrome (ACS) [15]. With
abdominal compliance defined as the measure of the ease of abdominal expansion, which is
determined by the elasticity of the abdominal wall and diaphragm, being the determining factor
explaining transmission of compartmental pressures from one compartment to another.
Fluid Balance. Daily fluid balance is the daily sum of all intakes and outputs, and
the cumulative fluid balance is the sum total of fluid accumulation over a set period of time [38,
54].
 Fluid overload. The percentage of fluid accumulation can be defined by dividing the
cumulative fluid balance in litre by the patient’s baseline body weight and multiplying by 100%.
Fluid overload is defined by a cut off value of 10% of fluid accumulation as this is associated with
worse outcomes [38, 55].
Fluid bolus. A rapid fluid infusion given as a bolus to correct hypotensive and
hypovolemic (septic or haemorrhagic) shock. It typically includes the infusion of at least 4 mL kg-
1 given over a maximum of 10 to 15 minutes.
Fluid challenge. A bolus of 100–200 mL given over 5–10 min with reassessment of
haemodynamic status to optimise tissue perfusion. This allows the construction of a so-called
Frank-Starling curve in order to assess the type of the curve and the position where the patient
is located on the curve. The CVP and pulmonary capillary wedge pressure (PCWP) are potentially
dangerous and useless to guide a fluid challenge [5, 13, 56, 57]. In the past, dynamic changes in
CVP (or PCWP) have been suggested but these may also not be useful [58, 59]. During a fluid
challenge, the 2—5 rule is classically followed for CVP and the 3—7 for PCWP. Baseline CVP
is measured and re-assessed after each bolus or each ten-minute period (as illustrated in Table
1).
Early adequate goal directed fluid management (EAFM). Most studies looking at goal
directed treatment def ine achieving the early goal as giving 25 to 50 mL kg-1 of fluids within the
first 6−8 hours of resuscitation in a case of septic or hypovolemic shock. However, others
have argued that such large volumes of fluid lead to ‘iatrogenic salt water drowning’ and have
proposed a more conservative strategy [13, 60].
Late Conservative Fluid Management (LCFM). Recent studies have shown that late
conservative fluid management, defined as two consecutive days of negative fluid balance within
the first week of ICU stay, is a strong and independent predictor of survival [21]. In contrast,
patients with persistent systemic inflammation maintain transcapillary albumin leakage and do not
reach the flow phase mounting up positive fluid balances.
Late Goal Directed Fluid Removal (LGFR). In some patients, more aggressive and active
fluid removal by means of diuretics and renal replacement therapy with net ultraf iltration is
needed. This is referred to as ‘de-resuscitation’.
Classification of fluid dynamics. Combining early adequate (EA) or early conservative
(EC) and late conservative (LC) or late liberal (LL) fluid management, four distinct groups can be
identified with regard to the dynamics of f luid management: EALC, EALL, ECLC, and ECLL.
These will be discussed further.
METHODS
search strategy and clinical questions
We searched MEDLINE, PubMed, EMBASE, Web of Science, The Cochrane Database,
clinical trials registries and bibliographies of included articles in order to update a previously
conducted systematic review and meta-analysis [61]. We sought to identify studies involving
critically ill patients that examined the association between a positive fluid balance and outcomes
after day 3 of ICU stay. We also sought to update a previously conducted systematic review
and meta-analysis the clinical questions of which were formulated according to the PICOD
(Patients, Interventions, Comparator, Outcome, Design) format [61]: Our PICOD clinical question
was: Does a management strategy in critically ill patients which attempts to achieve a fluid balance
in equilibrium or even negative (conservative fluid strategy) after day 3 (and within the first week)
result in a lower IAP and improved patient outcomes compared to a liberal fluid strategy?
article selection and data extraction
Two authors independently reviewed all titles and abstracts and selected full-text articles
for inclusion in the review. These two authors also abstracted the data on study design,
methodological quality, patient characteristics, fluid balance and the outcomes of interest. We
included studies where: 1) patients were critically ill or injured adults treated in an intensive care
unit (some of them also received surgery); 2) The intervention was any strategy or protocol
attempting to obtain a neutral or negative cumulative fluid balance after the third day of intensive
care; 3) the comparator group received a comparable strategy or protocol not attempting to obtain
negative fluid balance or equilibrium after the third day of intensive care allowing for a more
liberal fluid management strategy; 4) the primary outcomes were the incidence of IAH and
mortality; and 5) the study design was an RCT or observational study. Secondary outcomes
included cost, ICU utilisation, length of hospital or ICU stay, incidence of acute renal failure, ACS
and requirement for decompressive laparotomy. We also included previously conducted
systematic reviews and/or meta-analyses.
risk of bias assessment
We applied the Grades of Recommendation, Assessment, Development, and Evaluation
(GRADE) system to guide assessment of quality of evidence to address the clinical management
questions. The best consensus opinion was GRADED from high (A) to very low (D) and to help
determine the strength of recommendations from strong recommendations indicating that the
panel felt the overall desirable effects of the intervention clearly outweighed potential undesirable
effects, to weaker suggestions indicating that the balance of risks and benefits was less clear for
any intervention, to clear uncertainty.

statistical analysis
Continuous variables are expressed as mean ± SD (standard deviation) or as median (with
first and third quartile) according to whether they are normally distributed or not. The continuous
variables were compared using unpaired Student’s t-test for normally distributed variables and the
Mann-Whitney U for non-normally distributed variables. Random effects meta-analysis summary
results were calculated giving the average from the distribution (of treatment effects) across
studies. A P-value below 0.05 was considered as statistically significant. Statistical analysis was
done with SAS (version 9.1, SAS Institute, Cary, NC, USA) and SPSS (Windows version 17.0,
SPSS, Chicago, IL, USA). The meta-analysis and Forest plots were generated with Review
Manager 5 (Version 5.1. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration,
2011).
RESULTS
In total, we included 47 articles (surgical patients were studied in six, burns in three
and trauma in one, the other studies included mixed ICU (mainly medical) patients). We also
included one individual patient meta-analysis [62], 11 randomised controlled clinical trials
(RCTs) (of which four were blinded) [12, 30, 63-71], seven interventional studies [16, 72−77], 24
observational studies [21, 26, 28, 29, 31, 32, 49, 55, 78−93] and four case series [94−97].
Altogether, a total of 19902 critically ill patients were studied and in 20 studies the IAP was
measured (Table 2). In updating our previously conducted meta-analysis, we analysed the
following specific sub-questions:
Do non-survivors have a more positive fluid balance? A meta-analytic aproach was
adopted analysing the best available data abstracted from one individual patient meta-analysis [62],
nine uncontrolled prospective cohort studies [26, 28, 49, 77, 78, 82, 84, 89, 91], three uncontrolled
retrospective cohort studies [21, 31, 32], two retrospective non-randomised controlled cohort
studies [72, 79] and a retrospective review [86] of a randomised trial of a separate intervention
[68, 86] that considered fluid balance in relation to survival in critical illness. When compiled, the
data from a total of 5,445 patients from 17 studies showed that non-survivors (n= 2,609, 47.9%
mortality) had a more positive cumulative fluid balance by day 7 of their ICU stay compared
to survivors (6,982.6 ± 5,629 mL vs 2,449.1 ± 2,965.1 mL) (Fig. 1). The cumulative fluid
balance was on average 4,533.5 ± 3,626.7 mL more positive in non-survivors compared to
survivors (Figure 2). The collated f indings of these studies are provided in the Forest plot in Figure
3.
Does outcome improve with an intervention to limit fluid intake or lower fluid balance?
The compiled data from 15,947 patients enrolled in 28 studies [12, 16, 21, 28−32, 55, 63-68, 71,
72, 79−82, 84−87, 90, 91, 98] involving critically ill and peri-operative patients showed that
outcome was significantly improved when associated with a conservative fluid regimen (OR 0.42
[95% CI 0.32 to 0.55]), compared to non-conservative fluid management. This is illustrated in
the forest plot in Figure 4. In patients treated with a restrictive fluid regimen, mortality
decreased from 33.2% (2,596 deaths in 7,812 patients) to 24.7% (2,007 deaths in 8,135
patients, P < 0.0001). Actual data on cumulative fluid balance was available in 8,790 patients
from 16 studies [12, 16, 30, 55, 63−72, 90, 98, 99]: overall conservative treatment was associated
with a less positive fluid balance compared to a more liberal fluid strategy (2,131.7 ± 5,741.8 mL
vs 7,761 ± 7,391.9 mL) and the cumulative fluid balance was on average 5,629.3 ± 3,441.6 mL
less positive after one week of ICU stay (Figs 1, 2). The summary of findings of these studies is
given in Figure 5.
Do patients with IAH have a more positive fluid balance? Data was available from 1,517
patients obtained from one individual patient meta-analysis and seven cohort or case-controlled
studies [32, 62, 72, 78, 83, 88, 89, 100]. Meta-analysis of the pooled results revealed that the
597 patients with IAH (incidence being 39.4%) had a more positive f luid balance than those
without IAH (7,777.9 ± 3,803 mL vs 4,389.3 ± 1,996.4 mL) (Fig. 1). The cumulative fluid balance
after one week of ICU stay was on average 3,388.6 ± 2,324.2 mL more positive (Fig. 2). A
summary of the findings of these studies is given in Figure 6.
Does IAP improve with interventions acting on reducing fluid balance? Thirteen studies
investigated the effects of fluid removal (use of furosemide or renal replacement therapy with net
ultrafiltration) on IAP (Fig. 7). These were case studies or small series [70, 72−75, 77, 90, 92,
94−97]. A total fluid removal of 4,876.3 ± 4,178.5 mL resulted in a drop in IAP from 19.3 ± 9.1
to 11.5 ± 3.9 mm Hg (Fig. 8). A dose related effect was observed: the more negative the net fluid
balance or fluid removal, the greater the decrease in IAP (Fig. 9).
SUGGESTIONS FOR CLINICAL PRACTICE
Although the results of this meta-analysis are compelling, they are limited by indirectness and
the risk of bias given the inclusion of varying study designs and patient populations and the use
of many different interventions. After reviewing much of the above evidence, the World Society
of the Abdominal Compartment Syndrome suggested using a protocol to avoid a positive
cumulative fluid balance in critically ill patients, especially those with, or at risk of, IAH, after the
acute resuscitation has been completed and the inciting issues/source control have been addressed
(Grade 2C) [50].
We suggest a goal of a zero to negative fluid balance by day 3 and to keep the cumulative
fluid balance on day 7 as low as possible (Grade 2B). A vicious cycle leading to more f luid loading
and further IAP increase is illustrated in Figure 10, and this must be avoided. After reviewing the
limited evidence, we can only make a weak suggestion regarding the use of diuretics or renal
replacement therapy (in combi-nation with albumin) vs no intervention to mobilise fluids in
haemodynamically stable patients with IAH and a positive cumulative fluid balance after the acute
resuscitation has been completed and the inciting issues/source control have been addressed
(Grade 2D). The lack of consensus for this intervention underscores the uncertainity regarding its
role in managing the fluid balance and subsequently IAH, and the need for further studies.In
answer to the question ‘Why de-resuscitation?’: ‘Because fluid overload is independently related
to morbidity and mortality.

PATHOPHYSIOLOGY OF FLUID OVERLOAD

This section will address the question: ‘When to de-re-suscitate?’ As early as 1942, the
concept of a dual metabolic response to bodily injury was introduced. In direct response to initial
proinflammatory cytokines and stress hormones, the Ebb phase represents a distributive shock
characterised by arterial vasodilatation and transcapillary albumin leakage abating plasma oncotic
pressure. Arterial underfilling, microcirculatory dysfunction, and secondary interstitial oedema
lead to systemic hypoperfusion and regional impaired tissue use of oxygen. In this early stage of
shock, adequate fluid therapy comprises adequate goal directed f illing to prevent evolution to
multiple organ dysfunction syndrome (MODS). As compensatory neuroendocrine ref lexes and
potential renal dysfunction result in sodium and water retention, positive fluid balances are
inherent to the Ebb phase. Patients with higher severity of illness need more f luids to reach
cardiovascular optimisation. Therefore, at this point fluid balance may be considered a biomarker
of critical illness, as proposed by Bagshaw et al. [101]. Patients overcoming shock attain
homeostasis of proinflammatory and anti-inflammatory mediators classically within three days.
Subsequent haemodynamic stabilisation and restoration of plasma oncotic pressure set off the
Flow phase with resumption of diuresis and mobilisation of extravascular f luid resulting in
negative fluid balances.
When considering fluid administration, it is important to know when to start giving fluids
(what are the benefits of fluid administration), when to stop giving fluids (what are the risks of
ongoing fluid administration), when to start removing fluids (what are the benefits of fluid
removal), and when to stop fluid removal (what are the risks of removing too much fluid). The
literature shows that a negative fluid balance increases survival in patients with septic shock [31].
Patients admitted to the ICU who develop sepsis, respiratory failure, renal failure ARDS, IAH or
ACS all have a more positive cumulative fluid balance than those without organ failure [26, 27,
29, 102, 103]. Patients managed with a conservative fluid strategy also seem to have improved
lung function, shorter duration of mechanical ventilation and intensive care stay without
increasing non-pulmonary organ failure [30]. Monitoring is essential however, as any
measurement in the ICU will only be of value as long as it is accurate and reproducible, and no
measurement has ever improved survival, only a good protocol can do this. Vice versa a poor
treatment algorithm can result in potential harm to the patient [104, 105]. Patients who are in the
Ebb or Flow phase of shock have different clinical presentations and therefore different monitoring
needs (targets) and different treatment goals [25, 61].
Renal function in particular is strongly affected by fluid overload and IAH, and renal
interstitial oedema may impair renal function, even in the absence of IAH [14, 23, 91, 101, 106].
Therefore, fluid overload leading to IAH and associated renal dysfunction may counteract its own
resolution [107]. The adverse effects of fluid overload and interstitial oedema are numerous and
have an impact on all end organ functions, although some clinicians still believe that peripheral
oedema is only of cosmetic concern [108]. As adverse effects of fluid overload in states of
capillary leakage are particularly pronounced in the lungs, monitoring of EVLWI may offer a
valuable tool to guide f luid management in the critically ill. A high EVLWI indicates a state of
capillary leakage, associated with higher severity of illness and mortality [32, 72, 109, 110].
Previous studies correlated EVLWI with albumin extravasation in patients after multiple trauma
[111]. Responders to LCFM overcome the distributive shock and make a transition to the flow
phase [32]. On the other hand, nonresponders stay in the grip of the Ebb phase and progress to
GIPS, resulting in positive fluid balances, organ failure and death.
In this hypothesis, (change in) EVLWI has a prognostic value as a reflection of the extent
of capillary leakage, rather than as a quantification of lung function impairment by lung water [32,
47]. The recent observations may also have direct consequences regarding fluid management in
critically ill patients with IAH. Patients at risk for GIPS as assessed by CLI, IAP, changes in
EVLWI and fluid balance, require restrictive fluid strategies and even fluid removal guided by
extended haemodynamic monitoring including lung water measurements (late goal directed fluid
removal) [22, 112]. Previously, the application of EVLWI-guided fluid therapy led to improved
outcomes and lower positive fluid balances in states of capillary leakage [68]. To achieve
restrictive fluid management may necessitate a greater use of vasopressor therapy, resuscitation
with hyperoncotic solutions (e.g. albumin 20%) and early initiation of diuretics and renal
replacement therapy, although in the FACCT trial the conservative arm had a trend towards less
requirement for dialysis [30].
‘When should de-resuscitation begin?’: ‘De-resuscitation should be considered when fluid
overload and fluid accumulation negatively impact end-organ function, so de-resuscitation is
mandatory in a case of a positive cumulative f luid balance in combination with poor oxygenation
(P/F ratio < 200), increased capillary leak (high PVPI > 2.5 and EVLWI > 12 mL kg-1 PBW),
increased IAP (> 15 mm Hg) and low APP (< 50 mm Hg ), high CLI, etc.’
PRACTICAL APPROACH ‘How to de-resuscitate?’: Bedside measurement of
extravascular lung water (EVLWI) performed by trans-cardiopulmonary thermodilution allows
the estimation of the extent of capillary leak and fluid overload. Accordingly, EVLWI correlates
well with organ function and survival [49, 102, 109, 113]. Moreover, fluid management aimed
at EVLWI reduction results in a more negative fluid balance and improved outcomes [68]. In order
to achieve a negative fluid balance, previous prospective trials excluded patients with hypotension
and renal failure [30, 65, 68]. Recently in a study of 57 patients who were compared to 57 matched
controls, the effects of a restrictive fluid regimen with negative fluid balance using ‘PAL-
treatment’ were examined in mechanically ventilated patients with ALI presenting with severe
hypoxemia, increased EVLWI and IAP [72]. PAL-treatment combines high levels of positive end-
expiratory pressure (PEEP), small volume resuscitation with hyperoncotic albumin and fluid
removal with diuretics (Lasix®) or ultrafiltration during continuous renal replacement therapy
(CRRT). First, a 30-minute application of PEEP is titrated to counterbalance the effects of
increased IAP (best PEEP in cm H2O = IAP in mm Hg ). Next, hyperoncotic albumin (20%)
solution is administered by 200 mL boluses over 60 minutes twice on the f irst day and
subsequently titrated towards a serum albumin level of 30 g dL-1. Finally, 30 minutes after the
first albumin dose a furosemide infusion is initiated with an intravenous loading dose of 60 mg,
followed by a continuous infusion at 60 mg per hour for the first four hours and 5−20 mg per hour
thereafter, according to haemodynamic tolerance (Fig. 11). In anuric patients, CRRT can be added
with an ultrafiltration rate set in order to obtain a neutral to negative daily fluid balance. One week
of PAL-treatment had beneficial effects on EVLWI, IAP, organ function and vasopressor therapy,
and this resulted in a shorter duration of mechanical ventilation (faster weaning) and improved
28-day mortality (Fig. 12). Combining the results of two recent studies (n = 180), we found that
the group of patients treated with conservative initial and late fluid management had the best
outcome, followed by those who received initial adequate and late conservative fluid management
[32, 72]. Mortality was significantly increased in those patients who had received late liberal
fluid management (Fig. 14). This is in line with previous results by Murphy et al. [21].
DISCUSSION The results of this meta-analysis and systematic review of the available
evidence support the hypothesis that fluid overload is detrimental to patients and is associated with
increased morbidity and mortality.
We recently suggested a three-hit model of shock which we would like to extend to a four-
hit model in which we can recognise five distinct dynamic phases or stages of fluid administration
[32]: Resuscitation, Optimisation, Stabilisation, and Evacuation (ROSE), followed by a
potential risk of Hypoperfusion (Table 3, Fig. 15). Logically, these describe the f ive different
clinical phases of fluid therapy, occurring over the time course during which patients experience a
different impact on end-organ function (Fig. 15). Similar principles were recently also suggested
by others, confirming the need for a multicentre prospective trial with a bimodal approach using
late conservative fluid management after the initial early adequate goal-directed treatment in those
patients not transgressing spontaneously from the Ebb to the Flow phase [22, 24, 38, 40, 101, 114–
116]. We will discuss below the four-hit model of shock, each corresponding to a specific
treatment question.